AI Agents & MCP Server

Explore Granada runs a read-only Model Context Protocol (MCP) server for AI agents. There's no authentication and no API key. The server only reads; it never writes or changes anything. It doesn't expose any tools yet, so there's nothing to query through it today. What's live, and what isn't, is covered below.

This page is a static entry point for search engines and agent crawlers, nothing more. Publishing it doesn't make any agent discover the MCP server or connect to it on its own. An agent, or whoever is configuring it, still has to read the details below and set up the connection by hand.

Connect

• Endpoint: https://agents.exploregranada.org/mcp
• Transport: Streamable HTTP (JSON-RPC 2.0 over POST)
• Discovery document: https://agents.exploregranada.org/.well-known/mcp/server.json
• Full connection guide: https://agents.exploregranada.org/ — endpoint details, client config snippets, and notes on rate limits and logging

The _mcp.exploregranada.org DNS TXT record that some agent frameworks check first is not the MCP endpoint itself. It only points to the discovery document above.

Any MCP client that takes a JSON config

{
  "mcpServers": {
    "granada": { "type": "http", "url": "https://agents.exploregranada.org/mcp" }
  }
}

Tools

No tools are available on this server yet. A tools/list call currently returns an empty array. Check the connection guide above before assuming otherwise. This page gets updated once that changes.

Before you rely on it

• This is an early-access, experimental service. We're using it to learn what agents ask for before committing to a stable data feed.
• Requests are rate-limited per network address. Going over the limit returns an HTTP 429 with a Retry-After header.
• Calls are logged so we can see what agents request. IP addresses are never stored, and obvious personal data is stripped from logged arguments. Don't send personal data in tool arguments regardless.